Amazon, headquartered in Seattle, Washington, is a global leader in technology and an e-commerce powerhouse. Originally founded by Jeff Bezos in 1994 as a book marketplace, Amazon has since expanded to be one of the most influential and well-known online selling platforms throughout the global market.

To maximize sales, sellers on Amazon are responsible for performing many tasks. These include responding to customer inquiries, adjusting prices based on competitors, remembering when to stock new inventory, processing returns, and much more. Currently, these tasks are performed manually, requiring sellers to either take time out of their day or hire extra employees to accomplish these tasks.

Our solution, the Seller Agent Management Platform, addresses these challenges by enabling sellers to create customizable agents to automate tasks set by the seller. By offloading tasks, such as price adjustments or inventory stocking, to our advanced AI agents, sellers can save time while maintaining the same quality Amazon buyers have come to expect.

Users create agents through our web application, where they provide the agent with rules and settings on how to operate through our easy-to-use interface. Some of these settings include having the agent react to real-time events, assigning a task by providing a prompt, and specifying which capabilities the agent can or can’t have, as well as which actions need human approval before use. Sellers communicate with agents and approve agent actions using Slack, a messaging platform.

Our platform leverages AWS technologies such as Bedrock for LLM inference and Bedrock AgentCore for running the agents and accessing tools. The web application front end is built with NextJS, while the back end uses FastAPI, Amazon API Gateway, and Amazon Elastic Container Service.